The Effect of NOD2 Activation on TLR2-Mediated Cytokine Responses Is Dependent on Activation Dose and NOD2 Genotype

SHORT COMMUNICATION The effect of NOD2 activation on TLR2-mediated cytokine responses is dependent on activation dose and NOD2 genotype

MEA Borm1, AA van Bodegraven2, CJJ Mulder2, G Kraal1 and G Bouma1,2 1Department of Molecular Cell Biology and Immunology, Vrije Universiteit Medical Center, Amsterdam, The Netherlands and 2Department of Gastroenterology, Vrije Universiteit Medical Center, Amsterdam, The Netherlands

The mechanism by which mutations in NOD2 predispose to Crohn’s disease (CD) is incompletely understood. In mice, NOD2 has been found to function as a negative regulator of Toll-like receptor 2 (TLR2) signaling. In contrast, studies in humans so far showed no negative regulatory interaction between NOD2 and TLR2, and in fact suggest a synergistic effect between the two. Here, we show that this interaction is dose dependent. Adding low doses of muramyl dipeptide (MDP) to TLR2 primed monocytes results in a significant increase in cytokine production, whereas adding higher doses of MDP led to a striking downregulation of the responses. This downregulation by high-dose MDP does not occur in monocytes from NOD2-deficient patients. The inhibitory role of NOD2 at high concentrations of MDP implicates a safety mechanism to prevent exaggerated antibacterial immune responses in the gut to high or perpetuating bacterial load. This regulatory mechanism is lost in NOD2- deficient CD patients. Genes and Immunity (2008) 9, 274–278; doi:10.1038/gene.2008.9; published online 13 March 2008

Keywords: Crohn’s disease; NOD2; TLRs; immune activation

Mutations in the gene encoding nucleotide-binding peptidoglycan derivates).6 In mice, NOD2 has been oligomerization domain 2 (NOD2) are strongly asso- suggested to play a role as a negative regulator of ciated with an increased risk of developing Crohn’s TLR2 signaling, leading to excessive NF-kB dependent 1,2 disease (CD), a chronic inflammatory disorder of the IL-12 production associated with the strong Th1 response gastrointestinal tract. NOD2 is a member of the NOD- seen in murine colitis in NOD2 deficient mice.7,8 In leucine-rich repeat (LRR) protein family, which com- contrast, in human no indications for a negative prises of proteins that all consist of a C-terminal ligand regulatory interaction between NOD2 and TLR2 exist. recognition domain, a central NOD domain, and an Data in human on the simultaneous activation through N-terminal protein–protein interaction domain.3 Homozy- NOD2 and TLRs, including TLR2, show a positive gosity for mutations in the LRR of NOD2 results in a 20– synergistic effect, an effect that is lost in individuals 40 times higher risk of developing CD, whereas the risk bearing NOD2 mutations.9–12 This is accompanied by a of developing the disease is approximately four times loss of NF-kB signaling function leading to impaired higher in individuals that are heterozygotic.4 How the cytokine production after NOD2 stimulation, and to alterations in the NOD2 protein exert their effects on an impaired costimulatory effect of NOD2 signaling disease susceptibility is not completely clear, not in the on TLR-mediated cytokine production.9–12 This effect last place because this sensor of the bacterial peptido- appears contradictory to the elevated levels of NF-kB glycan component muramyl dipeptide (MDP) is ex- activation-dependent Th1 cytokines typically found in pressed in various cell types, including Paneth cells as the inflamed tissues from patients with CD.13,14 well as macrophages and dendritic cells. Furthermore, In the current study, we intended to bridge these activation of NOD2 results in activation of multiple seemingly contradictory findings. We hypothesized that pathways, including the nuclear factor (NF)-kB and the mode of interaction between NOD2 and TLR2 may mitogen-activated protein kinase pathways, leading to a be dose dependent and that part of the controversy may variety of responses.5,6 be related to differences in experimental design and Both NOD2 and Toll-like receptor 2 (TLR2) are stimulation dose. To test this hypothesis, we isolated activated by the same bacterial products (that is, monocytes from 5 healthy donors and 15 CD patients divided in three groups of 5 carrying 0, 1 or 2 mutant NOD2 alleles, respectively. All patients had inactive Correspondence: Dr G Bouma, Department of Gastroenterology, disease at the time of the study and were on minimum Vrije Universiteit Medical Center, De Boelelaan 1118, Amsterdam or no drug treatment. After isolation and an overnight 1081 HZ, The Netherlands. E-mail: [email protected] rest, cells were stimulated with low doses of either Received 21 January 2008; accepted 21 January 2008; published the synthetic TLR2 ligand Pam3CSK4 (Pam), or the online 13 March 2008 TLR4 ligand lipopolysaccharide (LPS). In addition, Effect of NOD2 on TLR2 responses in human monocytes MEA Borm et al 275 different concentrations of MDP (0, 1, 10, 25 Stimulation of cells with either LPS or Pam alone and 100 mgmlÀ1) were added and production of TNF-a, resulted in significant upregulation of cytokine produc- IL-6 and IL-12 was determined in the cell culture tion as compared to unstimulated cells. Stimulation of supernatant. cells with MDP alone resulted only in a small and

Figure 1 TNF-a production in relation to NOD2 genotype in monocytes stimulated with different TLR ligands and different doses of MDP. Monocytes were isolated from 50 ml of heparinized peripheral blood by density gradient centrifugation. Cells were cultured in triplicate (105 cells per 100 ml culture medium) and allowed to rest overnight. They were then stimulated for 3 h with pure synthetic TLR2 ligand À1 À1 (Pam3CSK4;50pgml ) and different concentrations of MDP (0, 1, 10, 25 and 100 mgml ). TNF-a was determined by enzyme-linked immunosorbent assay (ELISA). (A) TNF-a secretion after stimulation with the TLR2 ligand Pam3CSK4. Each graph includes five different donors stimulated with three different MDP doses (0 mgmlÀ1 (open bars); 100 mgmlÀ1 (black bars), whereas the gray bars represents the low MDP dose in the 1–25 mgmlÀ1 range that resulted in optimal cytokine secretion for that individual). CD, Crohn’s disease. Table 1a represent the absolute values of TNF-a secretion. (B)TNF-a secretion after stimulation of cells with the TLR4 ligand LPS (10 ng mlÀ1) and the different doses of MDP (also see Table 1b). The individual genotypes of the heterozygous and NOD2-deficient patients are provided in Supplementary Table 1.

Genes and Immunity Effect of NOD2 on TLR2 responses in human monocytes MEA Borm et al 276 Table 1a TNF-a responses in different groups of individuals according to NOD2 genotype in Pam3CSK4 stimulated monocytes in conjunction with different concentrations of MDP

TNF-a (in pg mlÀ1) TNF-a (in pg mlÀ1) after stimulation TNF-a (in pg mlÀ1) after stimulation with high without MDP (range) with low dose MDP (range); P-value dose MDP (range); P-values (compared to 0 mgmlÀ1 (compared to 0 mgmlÀ1 MDP) MDP and low-dose MDP, respectively)

NOD2 competent 865 (28–2761) 1556 (176–4073) 754 (85–1857) (at least one normal o0.0001 0.5/0.0002 allele; n ¼ 15) No mutations in 693 (28–2761) 1625 (176–4073) 666 (85–1857) NOD2 (n ¼ 10) 0.002 0.9/0.004 NOD2 heterozygous 1221 (428–1674) 1420 (513–2169) 932 (591–1349) (n ¼ 5) 0.06 0.3/0.3 NOD2 deficient 783 (351–1313) 914 (357–1991) 1029 (355–2104) (n ¼ 5) 0.6 0.2/0.4

Abbreviation: MDP, muramyl dipeptide; TNF-a, tumor necrosis factor-a. Bold values indicate statistically significant P-values.

Table 1b TNF-a responses in different groups of individuals according to NOD2 genotype in LPS stimulated monocytes in conjunction with different concentrations of MDP

TNF-a (in pg mlÀ1) TNF-a (in pg mlÀ1) after stimulation TNF-a (in pg/mlÀ1) after stimulation with without MDP (range) with low dose MDP (range); P-value high dose MDP (range); P-value (compared to 0 mgmlÀ1 (compared to 0 mgmlÀ1 MDP) MDP and low-dose MDP, respectively)

NOD2 competent 1319 (43–3063) 1910 (161–5862) 1920 (260–5582) (at least one normal 0.0009 0.003/0.8 allele; n ¼ 15) No mutations in 997 (43–3063) 1679 (161–5862) 1703 (260–5582) NOD2 (n ¼ 10) 0.01 0.004/0.7 NOD2 heterozygous 1964 (931–2642) 2375 (920–3495) 2354 (1041–3731) (n ¼ 5) 0.1 0.3/1 NOD2 deficient 2789 (454–5083) 3340 (499–6275) 3412 (523–6693) (n ¼ 5) 0.6 0.6/0.6

Abbreviations: LPS, lipopolysaccharide; MDP, muramyl dipeptide; TNF-a, tumor necrosis factor-a. Bold values indicate statistically significant P-values.

nonsignificant increase of cytokine expression in indivi- 100 mgmlÀ1 MDP resulted in a marked reduction of duals with at least one functional NOD2 allele. The TNF-a responses as compared to the low dose and magnitude of this response was 5–10 folds lower as at this concentration of MDP, TNF-a values returned compared to stimulation with the TLR ligands and to, or below the baseline level that was seen after independent of the dose (results not shown). As Pam stimulation alone (754 pg mlÀ1 (range 85–1857), expected, MDP stimulation of cells in NOD2-deficient P ¼ 0.0002 as compared with the low MDP dose). The individuals did not result in any upregulation of magnitude of this response was particularly evident in cytokine expression. individuals without any mutation in NOD2 and less Shown in Figure 1A and Table 1a is the TNF-a pronounced in heterozygous individuals, which is response of monocytes from different groups of CD indicative of a gene-dosage effect (Table 1a). patients and healthy controls after combined stimulation As expected, adding MDP (either low or high dose) with 50 ng mlÀ1 of the TLR2 ligand Pam and increasing did not result in upregulation of TNF-a in monocytes doses of MDP. Adding MDP in the dose range of from patients carrying mutations on both alleles (re- 1–25 mgmlÀ1 to Pam-stimulated monocytes resulted in a ferred to here as NOD2-deficient patients; Figure 1A and dose-dependent upregulation of TNF-a responses in 13 Table 1a). Thus, in sharp contrast to the biphasic out of 15 individuals (either patients or healthy controls) response seen in individuals carrying at least one carrying at least one wild-type NOD2 allele. The dose of NOD2 wild-type allele, no effect of MDP on the TLR2- MDP needed for maximum cytokine production differed induced response was seen in NOD2-deficient indivi- somewhat from person to person and was in the range duals. between 1 and 25 mgmlÀ1. For clarity of the graphs, we One could argue that the MDP concentration added is show only the dose that resulted in the highest cytokine fairly high, and that the effects seen in cytokine production for that particular individual (referred to as production are not due to an active inhibition of the low dose MDP, hereafter). Overall, the mean TNF-a TLR2 signaling, but due to toxic effects from the high response among these individuals increased from 865 MDP concentrations. The cells that received the high (range 28–2761) pg mlÀ1 after Pam stimulation alone to MDP dose however, remained viable even after pro- 1556 (range 176–4073) pg mlÀ1 after Pam þ low dose longed culture. In addition, as shown in Figure 1B and MDP, Po0.0001). Strikingly, adding a higher dose of Table 1b, the decrease in cytokine production seen after

Genes and Immunity Effect of NOD2 on TLR2 responses in human monocytes MEA Borm et al 277 Neg. Control 450 1500 300 400 100 g/ml MDP 350 300 1000 200 Pam CSK 250 3 4 200 150 500 100 Pam CSK +1 g/ml MDP

neg. control 3 4 Non Mutated

100 neg. control neg. control % compared to % compared to 50 % compared to 0 0 0 Pam3CSK4 +100 g/ml MDP p50 p65 c-Rel

125 450 125 400 100 350 100 300 75 250 75 50 200 50 150 neg. control neg. control 25 100 neg. control 25 % compared to Heterozygous % compared to 50 % compared to 0 0 0 p50 p65 c-Rel

200 200 125 100 75 100 100 50 neg. control neg. control

neg. control 25 % compared to % compared to % compared to NOD2 Deficient 0 0 0 p50 p65 c-Rel p50 p65 c-Rel Figure 2 Translocation of NF-kB subunits in relation to NOD2 genotype in monocytes. Monocytes were isolated from a nonmutated donor, a heterozygous donor and a NOD2-deficient donor. Monocytes were allowed to rest overnight, and then stimulated with MDP alone À1 À1 À1 À1 (100 mgml , dotted bar), Pam3CSK4 (50 pg ml , striped bar), Pam3CSK4 þ 1 mgml MDP (gray bar) or Pam3CSK4 þ 100 mgml MDP (black bar), or left unstimulated (negative control, open bar). After 20 min incubation, nuclear extracts were isolated using the Nuclear Extraction kit (Active Motif, Carlsbad, CA, USA), and the relative translocation of the NF-kB subunits p50, p65 and c-Rel was assessed by comparing the OD values in a NF-kB specific enzyme-linked immunosorbent assay (ELISA; Active Motif). adding the high dose of MDP in Pam-stimulated response was not seen after costimulation with LPS monocytes was not observed in cells costimulated with instead of Pam (see Supplementary Information file). the TLR4 ligand LPS instead of Pam. Taken together, To further substantiate these findings, we assessed these findings indicate that NOD2 signaling influences NF-kB activation in a small group of individuals. Monocytes TLR2 signaling in a biphasic fashion, first augmenting from a donor without NOD2 mutations, a NOD2 the response, and at a higher dose, diminishing the heterozygous donor and a NOD2-deficient donor were response. This response is TLR2 specific, independent stimulated with either Pam, Pam and low-dose MDP, from disease status and abrogated in NOD2-deficient and Pam with the high dose of MDP. As can be seen from individuals. Figure 2, the amount of translocated NF-kB p50 and With regard to IL-6, a similar response as seen for p65 subunits in the NOD2 competent donor, and to a TNF-a was found. Thus, an initial increase in IL-6 lesser extent in the heterozygous donor followed production was seen after addition of the low dose of a similar pattern as the cytokine responses, although it MDP to monocytes primed with Pam (from 698 pg mlÀ1 must be noted that this experiment was too small (range 100–2217) to 1592 pg mlÀ1 (range 518–5640), to draw definitive conclusions on the effect of MDP on P ¼ 0.001). Adding the high MDP dose resulted in a NF-kB translocation. This response was again not seen in significant lower IL-6 production (783 pg mlÀ1 (range the NOD2-deficient individual. Interestingly, this re- 109–2923); P ¼ 0.004 as compared to the low MDP dose). sponse was not reflected in c-Rel translocation. In fact, Similar to the results found for TNF-a, no effect of MDP c-Rel was hardly upregulated, suggesting that the p50 on IL-6 production was seen in monocytes from NOD2- and p65 units may be stronger induced by NOD2 deficient patients. Also, this biphasic response to MDP activation. was not seen in monocytes primed with LPS instead of The data presented here show that TLR2 signaling and Pam (see Supplementary Information file for the corre- NOD2 signaling in human monocytes are linked. In sponding graphs and tables). monocytes that are activated with TLR ligands and low For IL-12, a similar pattern was also found. Thus, IL- levels of the MDP, a synergistic effect is seen, correlating 12p40 increased from 8075 (range 1611–18611) pg mlÀ1 with earlier reports. However, when high doses of MDP after Pam alone to 13148 (range 2432–30648) pg mlÀ1 after are used this synergistic effect is lost in the case of TLR2 addition of the low dose MDP (Po0.0001) and returned signaling, but not in the case of TLR4. CD patients with to 8667 (range 546–22283) pg mlÀ1 after adding the high both NOD2 alleles mutated do not show this down- MDP dose (Po0.0001) in individuals carrying at least regulation, whereas a gene dosage effect is seen in one wild-type allele. For IL-12p70 the same pattern was patients with one allele affected. The importance of the found, although it must be noted that similar to the linkage between TLR2 and NOD2 signaling pathways, findings in mice, the IL-12p70 responses were generally which we show here for the fist time in human, may be low. Again, no measurable effect of MDP on IL-12p40 related to the fact that both TLR2 and NOD2 recognize and IL-12p70 production was found in monocytes closely related structures with MDP being a derivative of from NOD2-deficient patients. Similarly, this biphasic the TLR2 ligand peptidoglycan.5

Genes and Immunity Effect of NOD2 on TLR2 responses in human monocytes MEA Borm et al 278 The biphasic response seen with lower ligand stimula- variants with susceptibility to Crohn’s disease. Nature 2001; tion being synergistic, and only at high ligand levels 411: 599–603. leading to negative regulation, points to complex 3 Inohara N, Chamaillard M, McDonald C, Nunez G. NOD-LRR signaling pathways. The loss of regulatory function of proteins: role in host-microbial interactions and inflammatory the NOD2 molecule, as inferred from our experiments, disease. Annu Rev Biochem 2005; 74: 355–383. 4 Bouma G, Strober W. The immunological and genetic basis can explain the excessive Th1 responses seen in IBD patients and the strong linkage of the mutation of NOD2 of inflammatory bowel disease. Nat Rev Immunol 2003; 3: 521–533. and disease. It remains to be seen whether this function 5 Inohara N, Ogura Y, Fontalba A, Gutierrez O, Pons F, Crespo J is important at the initial stage of recognition of et al. Host recognition of bacterial muramyl dipeptide microbial pathogens or plays a role at later stages of mediated through NOD2: implications for Crohn’s disease. inflammation. The biphasic effects of NOD2 stimulation J Biol Chem 2003; 278: 5509–5512. on TLR2 signaling could also be reflected in the overall 6 Girardin SE, Boneca IG, Viala J, Chamaillard M, Labigne A, effects of NOD2 mutation on the pathogenesis of CD. Thomas G et al. NOD2 is a general sensor of peptidoglycan First, an increase in susceptibility to infection by through muramyl dipeptide (MDP) detection. J Biol Chem commensal bacteria could be caused by a reduced 2003; 278: 8869–8872. barrier function of the intestinal epithelium, due to 7 Watanabe T, Kitani A, Murray PJ, Strober W. NOD2 is a negative regulator of Toll-like receptor 2-mediated T helper impaired a-defensin production by Paneth cells,15 or a 16 type 1 responses. Nat Immunol 2004; 5: 800–808. reduced IL-1b responses to bacterial challenge 8 Watanabe T, Kitani A, Murray PJ, Wakatsuki Y, Fuss IJ, Strober combined with a less strong initial potentiating of W. Nucleotide binding oligomerization domain 2 deficiency TLR responses.9–12 This way NOD2 variants could lead leads to dysregulated TLR2 signaling and induction of to a weaker mucosal defense, leading to overgrowth of antigen-specific colitis. Immunity 2006; 25: 473–485. commensal bacteria. 9 van Heel DA, Ghosh S, Hunt KA, Mathew CG, Forbes A, Second, when there is an ongoing mucosal immune Jewell DP et al. Synergy between TLR9 and NOD2 innate response, and the negative regulation of NOD2 on immune responses is lost in genetic Crohn’s disease. Gut 2005; specifically TLR2 is lacking, this can result in an 54: 1553–1557. unwanted strong response.7,8 Without a proper down- 10 Netea MG, Ferwerda G, de Jong DJ, Jansen T, Jacobs L, Kramer M et al. Nucleotide-binding oligomerization domain-2 modulation of the immune response, this provides an modulates specific TLR pathways for the induction of activated immune setting for ongoing inflammation as cytokine release. J Immunol 2005; 174: 6518–6523. seen in CD. That the expression of NOD2 can be induced 11 Bonen DK, Ogura Y, Nicolae DL, Inohara N, Saab L, Tanabe T in T cells under the influence of pro-inflammatory factors et al. Crohn’s disease-associated NOD2 variants share a such as TNF-a,17 adds another level of complexity to the signaling defect in response to lipopolysaccharide and etiology of CD and suggests that NOD2 could also play a peptidoglycan. Gastroenterology 2003; 124: 140–146. role in later stages of the disease when the adaptive 12 van Heel DA, Ghosh S, Butler M, Hunt KA, Lundberg AM, immune system gets involved. Ahmad T et al. Muramyl dipeptide and toll-like receptor Clearly, NOD2 is a key player in innate immune sensitivity in NOD2-associated Crohn’s disease. Lancet 2005; 365 responses of many cell types. The lack of an initial strong : 1794–1796. 13 Monteleone G, Biancone L, Marasco R, Morrone G, Marasco response, combined with subsequent insufficient TLR2 O, Luzza F et al. Interleukin 12 is expressed and actively inhibition due to NOD2 defects could be an attractive released by Crohn’s disease intestinal lamina propria mono- mechanism that plays an important part in CD patho- nuclear cells. Gastroenterology 1997; 112: 1169–1178. genesis. To determine the exact role(s) of NOD2 in the 14 Fuss IJ, Neurath M, Boirivant M, Klein JS, de la Motte C, etiology of CD, it will be important to dissect the Strong SA et al. Disparate CD4+ lamina propria (LP) particular role of NOD2 and its interaction with TLR2 lymphokine secretion profiles in inflammatory bowel disease. in the various cell types involved in the intestinal Crohn’s disease LP cells manifest increased secretion of IFN- homeostasis, such as epithelial Paneth cells and hemo- gamma, whereas ulcerative colitis LP cells manifest increased poietic immune cells. secretion of IL-5. J Immunol 1996; 157: 1261–1270. 15 Lala S, Ogura Y, Osborne C, Hor SY, Bromfield A, Davies S et al. Crohn’s disease and the NOD2 gene: a role for paneth cells. Gastroenterology 2003; 125: 47–57. 16 Li J, Moran T, Swanson E, Julian C, Harris J, Bonen DK et al. References Regulation of IL-8 and IL-1beta expression in Crohn’s disease associated NOD2/CARD15 mutations. Hum Mol Genet 2004; 1 Ogura Y, Bonen DK, Inohara N, Nicolae DL, Chen FF, Ramos 13: 1715–1725. R et al. A frameshift mutation in NOD2 associated with 17 Gutierrez O, Pipaon C, Inohara N, Fontalba A, Ogura Y, susceptibility to Crohn’s disease. Nature 2001; 411: 603–606. Prosper F et al. Induction of NOD2 in myelomonocytic and 2 Hugot JP, Chamaillard M, Zouali H, Lesage S, Cezard JP, intestinal epithelial cells via nuclear factor-kB. J Biol Chem Belaiche J et al. Association of NOD2 leucine-rich repeat 2002; 277: 41701–41705.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity